API ReferenceLocal APIs

useVAD

React hook — real-time mic voice-activity detection (Silero).

Auto-generated from JSDoc in packages/react/src/local/useVAD.ts. Do not edit — run npm run docs:api.

useVAD

function

function useVAD(opts: UseVADOptions): UseVAD

UseVAD

interface

interface UseVAD

`status: VADStatus`

`error?: Error`

`start: () => Promise<void>`

Request mic permission, load the model, begin listening.

`stop: () => void`

Stop listening (keeps the loaded model for a fast restart).

UseVADOptions

interface

interface UseVADOptions

`onSpeechEnd?: (audio: Float32Array) => void`

Called with the 16 kHz Float32 audio of each completed speech segment.

`onSpeechStart?: () => void`

`baseAssetPath?: string`

Override the CDN paths for the model + onnx wasm if self-hosting.

`onnxWASMBasePath?: string`

VADStatus

type alias

type VADStatus = idle | loading | listening | speaking | error

useLocalTTS

React hook — on-device text-to-speech (Kokoro) over WebGPU.

useTranscribe

React hook — speech-to-text (Whisper) over WebGPU.

On this page

useVAD UseVAD status: VADStatuserror?: Errorstart: () => Promise<void>stop: () => voidUseVADOptions onSpeechEnd?: (audio: Float32Array) => voidonSpeechStart?: () => voidbaseAssetPath?: stringonnxWASMBasePath?: stringVADStatus

Valora is local-first

No API key, no server — everything in this doc runs on-device.